Closed downor closed 6 years ago
Right, that regression was the reason why this patch was reverted from Mesa upstream branch.
Thanks for this however, one question to tpalli @tpalli So I guess there is some contradiction here in this situation. According to the commit msg of original patch, GL_HALF_FLOAT may be needed to be supported by ReadPixels. Doesn't this mean conformance test needs to be modified to pass this? What would be the long term solution for this?
@downor Yes, I'm pretty sure the Android CTS test has a bug since none of the Khronos CTS or dEQP test uses that format for reading. It would make sense to ask Google about this or maybe file a bug (if that is possible) against this test-suite.
…readpixels"""
This reverts commit 8aa5c8b310f6455ad82a52fc4ff082a7b4ca9a25.
Original patch, "mesa: add GL_HALF_FLOAT as supported type to readpixels" needs to be reverted since it causes a regression with "KHR-GLES3.packed_pixels.* tests".